json-parser.c

   1 /*
   2  * JSON Parser
   3  *
   4  * Copyright IBM, Corp. 2009
   5  *
   6  * Authors:
   7  *  Anthony Liguori   <aliguori@us.ibm.com>
   8  *
   9  * This work is licensed under the terms of the GNU LGPL, version 2.1 or later.
  10  * See the COPYING.LIB file in the top-level directory.
  11  *
  12  */
  13
  14 #include <stdbool.h>
  15 #include <stdarg.h>
  16
  17 #include "qemu-common.h"
  18 #include "qstring.h"
  19 #include "qint.h"
  20 #include "qdict.h"
  21 #include "qlist.h"
  22 #include "qfloat.h"
  23 #include "qbool.h"
  24 #include "json-parser.h"
  25 #include "json-lexer.h"
  26
  27 typedef struct JSONParserContext
  28 {
  29 } JSONParserContext;
  30
  31 #define BUG_ON(cond) assert(!(cond))
  32
  33 /**
  34  * TODO
  35  *
  36  * 0) make errors meaningful again
  37  * 1) add geometry information to tokens
  38  * 3) should we return a parsed size?
  39  * 4) deal with premature EOI
  40  */
  41
  42 static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap);
  43
  44 /**
  45  * Token manipulators
  46  *
  47  * tokens are dictionaries that contain a type, a string value, and geometry information
  48  * about a token identified by the lexer.  These are routines that make working with
  49  * these objects a bit easier.
  50  */
  51 static const char *token_get_value(QObject *obj)
  52 {
  53     return qdict_get_str(qobject_to_qdict(obj), "token");
  54 }
  55
  56 static JSONTokenType token_get_type(QObject *obj)
  57 {
  58     return qdict_get_int(qobject_to_qdict(obj), "type");
  59 }
  60
  61 static int token_is_operator(QObject *obj, char op)
  62 {
  63     const char *val;
  64
  65     if (token_get_type(obj) != JSON_OPERATOR) {
  66         return 0;
  67     }
  68
  69     val = token_get_value(obj);
  70
  71     return (val[0] == op) && (val[1] == 0);
  72 }
  73
  74 static int token_is_keyword(QObject *obj, const char *value)
  75 {
  76     if (token_get_type(obj) != JSON_KEYWORD) {
  77         return 0;
  78     }
  79
  80     return strcmp(token_get_value(obj), value) == 0;
  81 }
  82
  83 static int token_is_escape(QObject *obj, const char *value)
  84 {
  85     if (token_get_type(obj) != JSON_ESCAPE) {
  86         return 0;
  87     }
  88
  89     return (strcmp(token_get_value(obj), value) == 0);
  90 }
  91
  92 /**
  93  * Error handler
  94  */
  95 static void parse_error(JSONParserContext *ctxt, QObject *token, const char *msg, ...)
  96 {
  97     va_list ap;
  98     va_start(ap, msg);
  99     fprintf(stderr, "parse error: ");
 100     vfprintf(stderr, msg, ap);
 101     fprintf(stderr, "\n");
 102     va_end(ap);
 103 }
 104
 105 /**
 106  * String helpers
 107  *
 108  * These helpers are used to unescape strings.
 109  */
 110 static void wchar_to_utf8(uint16_t wchar, char *buffer, size_t buffer_length)
 111 {
 112     if (wchar <= 0x007F) {
 113         BUG_ON(buffer_length < 2);
 114
 115         buffer[0] = wchar & 0x7F;
 116         buffer[1] = 0;
 117     } else if (wchar <= 0x07FF) {
 118         BUG_ON(buffer_length < 3);
 119
 120         buffer[0] = 0xC0 | ((wchar >> 6) & 0x1F);
 121         buffer[1] = 0x80 | (wchar & 0x3F);
 122         buffer[2] = 0;
 123     } else {
 124         BUG_ON(buffer_length < 4);
 125
 126         buffer[0] = 0xE0 | ((wchar >> 12) & 0x0F);
 127         buffer[1] = 0x80 | ((wchar >> 6) & 0x3F);
 128         buffer[2] = 0x80 | (wchar & 0x3F);
 129         buffer[3] = 0;
 130     }
 131 }
 132
 133 static int hex2decimal(char ch)
 134 {
 135     if (ch >= '0' && ch <= '9') {
 136         return (ch - '0');
 137     } else if (ch >= 'a' && ch <= 'f') {
 138         return 10 + (ch - 'a');
 139     } else if (ch >= 'A' && ch <= 'F') {
 140         return 10 + (ch - 'A');
 141     }
 142
 143     return -1;
 144 }
 145
 146 /**
 147  * parse_string(): Parse a json string and return a QObject
 148  *
 149  *  string
 150  *      ""
 151  *      " chars "
 152  *  chars
 153  *      char
 154  *      char chars
 155  *  char
 156  *      any-Unicode-character-
 157  *          except-"-or-\-or-
 158  *          control-character
 159  *      \"
 160  *      \\
 161  *      \/
 162  *      \b
 163  *      \f
 164  *      \n
 165  *      \r
 166  *      \t
 167  *      \u four-hex-digits
 168  */
 169 static QString *qstring_from_escaped_str(JSONParserContext *ctxt, QObject *token)
 170 {
 171     const char *ptr = token_get_value(token);
 172     QString *str;
 173     int double_quote = 1;
 174
 175     if (*ptr == '"') {
 176         double_quote = 1;
 177     } else {
 178         double_quote = 0;
 179     }
 180     ptr++;
 181
 182     str = qstring_new();
 183     while (*ptr &&
 184            ((double_quote && *ptr != '"') || (!double_quote && *ptr != '\''))) {
 185         if (*ptr == '\\') {
 186             ptr++;
 187
 188             switch (*ptr) {
 189             case '"':
 190                 qstring_append(str, "\"");
 191                 ptr++;
 192                 break;
 193             case '\'':
 194                 qstring_append(str, "'");
 195                 ptr++;
 196                 break;
 197             case '\\':
 198                 qstring_append(str, "\\");
 199                 ptr++;
 200                 break;
 201             case '/':
 202                 qstring_append(str, "/");
 203                 ptr++;
 204                 break;
 205             case 'b':
 206                 qstring_append(str, "\b");
 207                 ptr++;
 208                 break;
 209             case 'f':
 210                 qstring_append(str, "\f");
 211                 ptr++;
 212                 break;
 213             case 'n':
 214                 qstring_append(str, "\n");
 215                 ptr++;
 216                 break;
 217             case 'r':
 218                 qstring_append(str, "\r");
 219                 ptr++;
 220                 break;
 221             case 't':
 222                 qstring_append(str, "\t");
 223                 ptr++;
 224                 break;
 225             case 'u': {
 226                 uint16_t unicode_char = 0;
 227                 char utf8_char[4];
 228                 int i = 0;
 229
 230                 ptr++;
 231
 232                 for (i = 0; i < 4; i++) {
 233                     if (qemu_isxdigit(*ptr)) {
 234                         unicode_char |= hex2decimal(*ptr) << ((3 - i) * 4);
 235                     } else {
 236                         parse_error(ctxt, token,
 237                                     "invalid hex escape sequence in string");
 238                         goto out;
 239                     }
 240                     ptr++;
 241                 }
 242
 243                 wchar_to_utf8(unicode_char, utf8_char, sizeof(utf8_char));
 244                 qstring_append(str, utf8_char);
 245             }   break;
 246             default:
 247                 parse_error(ctxt, token, "invalid escape sequence in string");
 248                 goto out;
 249             }
 250         } else {
 251             char dummy[2];
 252
 253             dummy[0] = *ptr++;
 254             dummy[1] = 0;
 255
 256             qstring_append(str, dummy);
 257         }
 258     }
 259
 260     return str;
 261
 262 out:
 263     QDECREF(str);
 264     return NULL;
 265 }
 266
 267 /**
 268  * Parsing rules
 269  */
 270 static int parse_pair(JSONParserContext *ctxt, QDict *dict, QList **tokens, va_list *ap)
 271 {
 272     QObject *key, *token = NULL, *value, *peek;
 273     QList *working = qlist_copy(*tokens);
 274
 275     peek = qlist_peek(working);
 276     key = parse_value(ctxt, &working, ap);
 277     if (!key || qobject_type(key) != QTYPE_QSTRING) {
 278         parse_error(ctxt, peek, "key is not a string in object");
 279         goto out;
 280     }
 281
 282     token = qlist_pop(working);
 283     if (!token_is_operator(token, ':')) {
 284         parse_error(ctxt, token, "missing : in object pair");
 285         goto out;
 286     }
 287
 288     value = parse_value(ctxt, &working, ap);
 289     if (value == NULL) {
 290         parse_error(ctxt, token, "Missing value in dict");
 291         goto out;
 292     }
 293
 294     qdict_put_obj(dict, qstring_get_str(qobject_to_qstring(key)), value);
 295
 296     qobject_decref(token);
 297     qobject_decref(key);
 298     QDECREF(*tokens);
 299     *tokens = working;
 300
 301     return 0;
 302
 303 out:
 304     qobject_decref(token);
 305     qobject_decref(key);
 306     QDECREF(working);
 307
 308     return -1;
 309 }
 310
 311 static QObject *parse_object(JSONParserContext *ctxt, QList **tokens, va_list *ap)
 312 {
 313     QDict *dict = NULL;
 314     QObject *token, *peek;
 315     QList *working = qlist_copy(*tokens);
 316
 317     token = qlist_pop(working);
 318     if (!token_is_operator(token, '{')) {
 319         goto out;
 320     }
 321     qobject_decref(token);
 322     token = NULL;
 323
 324     dict = qdict_new();
 325
 326     peek = qlist_peek(working);
 327     if (!token_is_operator(peek, '}')) {
 328         if (parse_pair(ctxt, dict, &working, ap) == -1) {
 329             goto out;
 330         }
 331
 332         token = qlist_pop(working);
 333         while (!token_is_operator(token, '}')) {
 334             if (!token_is_operator(token, ',')) {
 335                 parse_error(ctxt, token, "expected separator in dict");
 336                 goto out;
 337             }
 338             qobject_decref(token);
 339             token = NULL;
 340
 341             if (parse_pair(ctxt, dict, &working, ap) == -1) {
 342                 goto out;
 343             }
 344
 345             token = qlist_pop(working);
 346         }
 347         qobject_decref(token);
 348         token = NULL;
 349     } else {
 350         token = qlist_pop(working);
 351         qobject_decref(token);
 352         token = NULL;
 353     }
 354
 355     QDECREF(*tokens);
 356     *tokens = working;
 357
 358     return QOBJECT(dict);
 359
 360 out:
 361     qobject_decref(token);
 362     QDECREF(working);
 363     QDECREF(dict);
 364     return NULL;
 365 }
 366
 367 static QObject *parse_array(JSONParserContext *ctxt, QList **tokens, va_list *ap)
 368 {
 369     QList *list = NULL;
 370     QObject *token, *peek;
 371     QList *working = qlist_copy(*tokens);
 372
 373     token = qlist_pop(working);
 374     if (!token_is_operator(token, '[')) {
 375         goto out;
 376     }
 377     qobject_decref(token);
 378     token = NULL;
 379
 380     list = qlist_new();
 381
 382     peek = qlist_peek(working);
 383     if (!token_is_operator(peek, ']')) {
 384         QObject *obj;
 385
 386         obj = parse_value(ctxt, &working, ap);
 387         if (obj == NULL) {
 388             parse_error(ctxt, token, "expecting value");
 389             goto out;
 390         }
 391
 392         qlist_append_obj(list, obj);
 393
 394         token = qlist_pop(working);
 395         while (!token_is_operator(token, ']')) {
 396             if (!token_is_operator(token, ',')) {
 397                 parse_error(ctxt, token, "expected separator in list");
 398                 goto out;
 399             }
 400
 401             qobject_decref(token);
 402             token = NULL;
 403
 404             obj = parse_value(ctxt, &working, ap);
 405             if (obj == NULL) {
 406                 parse_error(ctxt, token, "expecting value");
 407                 goto out;
 408             }
 409
 410             qlist_append_obj(list, obj);
 411
 412             token = qlist_pop(working);
 413         }
 414
 415         qobject_decref(token);
 416         token = NULL;
 417     } else {
 418         token = qlist_pop(working);
 419         qobject_decref(token);
 420         token = NULL;
 421     }
 422
 423     QDECREF(*tokens);
 424     *tokens = working;
 425
 426     return QOBJECT(list);
 427
 428 out:
 429     qobject_decref(token);
 430     QDECREF(working);
 431     QDECREF(list);
 432     return NULL;
 433 }
 434
 435 static QObject *parse_keyword(JSONParserContext *ctxt, QList **tokens)
 436 {
 437     QObject *token, *ret;
 438     QList *working = qlist_copy(*tokens);
 439
 440     token = qlist_pop(working);
 441
 442     if (token_get_type(token) != JSON_KEYWORD) {
 443         goto out;
 444     }
 445
 446     if (token_is_keyword(token, "true")) {
 447         ret = QOBJECT(qbool_from_int(true));
 448     } else if (token_is_keyword(token, "false")) {
 449         ret = QOBJECT(qbool_from_int(false));
 450     } else {
 451         parse_error(ctxt, token, "invalid keyword `%s'", token_get_value(token));
 452         goto out;
 453     }
 454
 455     qobject_decref(token);
 456     QDECREF(*tokens);
 457     *tokens = working;
 458
 459     return ret;
 460
 461 out:
 462     qobject_decref(token);
 463     QDECREF(working);
 464
 465     return NULL;
 466 }
 467
 468 static QObject *parse_escape(JSONParserContext *ctxt, QList **tokens, va_list *ap)
 469 {
 470     QObject *token = NULL, *obj;
 471     QList *working = qlist_copy(*tokens);
 472
 473     if (ap == NULL) {
 474         goto out;
 475     }
 476
 477     token = qlist_pop(working);
 478
 479     if (token_is_escape(token, "%p")) {
 480         obj = va_arg(*ap, QObject *);
 481     } else if (token_is_escape(token, "%i")) {
 482         obj = QOBJECT(qbool_from_int(va_arg(*ap, int)));
 483     } else if (token_is_escape(token, "%d")) {
 484         obj = QOBJECT(qint_from_int(va_arg(*ap, int)));
 485     } else if (token_is_escape(token, "%ld")) {
 486         obj = QOBJECT(qint_from_int(va_arg(*ap, long)));
 487     } else if (token_is_escape(token, "%lld") ||
 488                token_is_escape(token, "%I64d")) {
 489         obj = QOBJECT(qint_from_int(va_arg(*ap, long long)));
 490     } else if (token_is_escape(token, "%s")) {
 491         obj = QOBJECT(qstring_from_str(va_arg(*ap, const char *)));
 492     } else if (token_is_escape(token, "%f")) {
 493         obj = QOBJECT(qfloat_from_double(va_arg(*ap, double)));
 494     } else {
 495         goto out;
 496     }
 497
 498     qobject_decref(token);
 499     QDECREF(*tokens);
 500     *tokens = working;
 501
 502     return obj;
 503
 504 out:
 505     qobject_decref(token);
 506     QDECREF(working);
 507
 508     return NULL;
 509 }
 510
 511 static QObject *parse_literal(JSONParserContext *ctxt, QList **tokens)
 512 {
 513     QObject *token, *obj;
 514     QList *working = qlist_copy(*tokens);
 515
 516     token = qlist_pop(working);
 517     switch (token_get_type(token)) {
 518     case JSON_STRING:
 519         obj = QOBJECT(qstring_from_escaped_str(ctxt, token));
 520         break;
 521     case JSON_INTEGER:
 522         obj = QOBJECT(qint_from_int(strtoll(token_get_value(token), NULL, 10)));
 523         break;
 524     case JSON_FLOAT:
 525         /* FIXME dependent on locale */
 526         obj = QOBJECT(qfloat_from_double(strtod(token_get_value(token), NULL)));
 527         break;
 528     default:
 529         goto out;
 530     }
 531
 532     qobject_decref(token);
 533     QDECREF(*tokens);
 534     *tokens = working;
 535
 536     return obj;
 537
 538 out:
 539     qobject_decref(token);
 540     QDECREF(working);
 541
 542     return NULL;
 543 }
 544
 545 static QObject *parse_value(JSONParserContext *ctxt, QList **tokens, va_list *ap)
 546 {
 547     QObject *obj;
 548
 549     obj = parse_object(ctxt, tokens, ap);
 550     if (obj == NULL) {
 551         obj = parse_array(ctxt, tokens, ap);
 552     }
 553     if (obj == NULL) {
 554         obj = parse_escape(ctxt, tokens, ap);
 555     }
 556     if (obj == NULL) {
 557         obj = parse_keyword(ctxt, tokens);
 558     }
 559     if (obj == NULL) {
 560         obj = parse_literal(ctxt, tokens);
 561     }
 562
 563     return obj;
 564 }
 565
 566 QObject *json_parser_parse(QList *tokens, va_list *ap)
 567 {
 568     JSONParserContext ctxt = {};
 569     QList *working = qlist_copy(tokens);
 570     QObject *result;
 571
 572     result = parse_value(&ctxt, &working, ap);
 573
 574     QDECREF(working);
 575
 576     return result;
 577 }